The shared axis that made baccarat’s swings run one way
Showing the misleading chart
A table-games review slide we drew puts four Nevada games on one ruler — monthly statewide hold for Twenty-One, craps, roulette and baccarat, Sep 2024 to Aug 2026, every month the Gaming Control Board printed — and concludes that baccarat’s swings only run downward: its best month tops out at 20%, level with roulette’s, while its worst falls to 8.54%. The shared axis stops at 20% and the data does not. Nine bars ran past it and were drawn exactly 20% tall, among them January 2025, when baccarat held 26.47% — the highest monthly hold any of the four games has posted in a record that runs back to 2004. Run the ruler to 28% and baccarat’s swings run both ways.
01The claim
A monthly hold review slide of the kind a table-games desk might circulate, which we drew from public numbers: four small multiples — Twenty-One, craps, roulette and baccarat — each showing Nevada’s statewide hold for the 24 months from September 2024 to August 2026. Hold is the casino’s win as a share of the drop, and every bar is the figure the Gaming Control Board printed for that game and month. The slide does a lot right: the bars start at zero, nothing is smoothed or left out, and all four panels share one fixed 0–20% axis so the games can be compared at a glance. Read that way, roulette and baccarat top out at exactly the same hold, 20.00%, while baccarat alone drops as low as 8.54% — so baccarat’s volatility looks like pure downside, and the slide recommends budgeting it as the riskier game and steering promotion toward roulette.
02The trick
The ceiling is a side effect of a good habit. Small multiples should share one axis, and this one does — but its 20% maximum looks as if it was set by the quiet panels, where Twenty-One never passed 14.99% and craps never passed 18.88%, and then applied to two games that regularly go past it. Nine bars did: seven of roulette’s and two of baccarat’s, each drawn exactly 20% tall, because a fixed maximum leaves a bar nowhere else to go. Most clipping stories are about a lost peak. This one is about spread. The ceiling cuts off only the top of each game’s range and leaves the bottom alone, so a game whose swings run both ways looks lopsided: baccarat’s true range of 17.93 points is drawn as 11.46, and its highest month — January 2025, 26.47%, the highest monthly hold any of the four games has posted in a record that runs back to 2004 — comes out the same height as roulette’s 20.19% that month. Even the count of bars at the top cannot be read off the panel: two of roulette’s unclipped months, 19.81% and 19.90%, sit within a hair of the seven that were clipped. The read-out’s “best month 20.00%” is the height of the frame, and the “downside only” conclusion is the shape of the ceiling.
03The fix
Keep the shared ruler and run it past the data. On one zero-based axis to 28%, with a dashed line where the old 20% maximum was and the overflow hatched, Twenty-One and craps look exactly as they did — the ceiling never touched them — while roulette pokes through seven times, peaking at 23.03% in August 2025, and baccarat’s January 2025 stands at 26.47%, 6.28 points above roulette’s for the same month. Altogether the ceiling had flattened 16.37 points of hold: 9.20 of roulette’s and 7.17 of baccarat’s, 6.47 of them in that one January. Baccarat’s swings turn out to run both ways. Over these 24 months it averaged 16.21%, finished 13 months above that, 10 below and one level with it, and the gap is larger on the high side (+10.26 points in January 2025) than the low side (−7.67 in September 2025). Its true range is 17.93 points against roulette’s 11.06; the slide drew them as 11.46 and 8.03. That is what makes baccarat baccarat: a statewide month rests on far fewer, far larger wagers than a month of Twenty-One, so a few tables’ luck moves it a long way in either direction. Two habits stop a chart like this. First, when panels share an axis, set the maximum from the largest value any panel reaches, not from the panel you looked at first. Where a fixed maximum really is needed, say to keep a tile comparable from month to month, make the overflow obvious: hatch or arrow the part that crosses it, print the value, and say in the caption how many marks went over. Second, never compute a “best” or “peak” from the drawing. A peak that equals the axis maximum is telling you about the chart’s settings, not about the data.